Leveraging Turbine-Level Data for Improved Probabilistic Wind Power Forecasting

Abstract

This paper describes two methods for creating improved probabilistic wind power forecasts through the use of turbine-level data. The first is a feature engineering approach whereby deterministic power forecasts from the turbine level are used as explanatory variables in a wind farm level forecasting model. The second is a novel bottom-up hierarchical approach where the wind farm forecast is inferred from the joint predictive distribution of the power output from individual turbines. Notably, the latter produces probabilistic forecasts that are coherent across both turbine and farm levels, which the former does not. The methods are tested at two utility scale wind farms and are shown to provide consistent improvements of up to 5%, in terms of continuous ranked probability score compared to the best performing state-of-the-art benchmark model. The bottom-up hierarchical approach provides greater improvement at the site characterized by a complex layout and terrain, while both approaches perform similarly at the second location. We show that there is a clear benefit in leveraging readily available turbine-level information for wind power forecasting.

ProbCast: Open-source Production, Evaluation and Visualisation of Probabilistic Forecasts

“…It provides a data structure, MultiQR, for storing tables of multiple predictive quantiles. It has been developed with energy applications in mind, for example [30], [31], but the functionality it provides is more generally applicable.…”

Statistical post‐processing of turbulence‐resolving weather forecasts for offshore wind power forecasting

“…For instance, the engineered features such as the rolling variance, which quantify the variability of the signal, could be more valuable in probabilistic forecasting for modelling the upper and lower ends of the distribution via quantile regression . Understandably, a hierarchical model where each turbine is used to generate a consistent wind farm forecast could be an optimal way of using the high spatial content of the data . Additionally, a more in depth‐study focused on extracting value from the temporal content of the wind forecast signal such as deep‐learning, instantaneous frequency transforms, or wavelet decomposition techniques should be explored.…”

“…27,45 Understandably, a hierarchical model where each turbine is used to generate a consistent wind farm forecast could be an optimal way of using the high spatial content of the data. 46 Additionally, a more in depth-study focused on extracting value from the temporal content of the wind forecast signal such as deep-learning, 47 instantaneous frequency transforms, 44 or wavelet decomposition 48 techniques should be explored.…”

Quantile Combination for the EEM20 Wind Power Forecasting Competition

“…While wind power forecasting is relatively mature and many commercial offerings exist today, research is ongoing to improve accuracy and forecast further ahead [2]. Notably, recent advances have come from incorporating new sources of data, such as girds of Numerical Weather Predictions (NWP) [3], turbine-level data [4], remote sensing [5], [6], and advances in NWP [7], enabled by contemporary data science.…”